offence-defence theory: towards a definitive understanding€¦ · pp. 108–146; stephen van...

Offence-defence Theory: Towardsa Definitive Understanding

Tang Shiping*

Offence-defence theory (ODT) dominates the understanding of many secur-

ity and international relations (IR) theorists of the role in international

politics that military technology plays. ODT has produced some of the

most cited works in realism literature,1 as evident in the many works

which—implicitly and explicitly—rely on ODT to propel their arguments.2

*Corresponding author. Email: [email protected]

Tang Shiping is Professor at the School of International Relations and Public Affairs(SIRPA), Fudan University, Shanghai, China. Prior to his current appointment,Shiping was Senior Fellow at the S. Rajaratnam School of International Studies,Nanyang Technological University, Singapore, where this article was finished. He alsothanks Taylor Fravel, Evan Montgomery, and Jack Snyder for helpful comments on anearlier draft. Beatrice Bieger provided outstanding research assistance. The usualdisclaimer applies.

1 George H. Quester, Offense and Defense in the International System (New York, N.Y.:John Wiley and Sons, 1977); Robert Jervis ‘‘Cooperation under the security dilemma,’’World Politics, Vol. 30, No. 2 (1978), pp. 167–214; Jack Snyder, ‘‘Civil-Military Relationsand the Cult of the Offensive, 1914 and 1984,’’ International Security, Vol. 9, No. 1 (1984),pp. 108–146; Stephen Van Evera, ‘‘The Cult of the Offensive and the Origins of the FirstWorld War,’’ International Security, Vol. 9, No. 1 (1984), pp. 58–107; Stephen Van Evera,Causes of War: Power and the Roots of Conflict (Ithaca, N.Y.: Cornell University Press,1999); and Charles L. Glaser and Chaim Kaufmann, ‘‘What Is the Offense-DefenseBalance and How Can We Measure It?’’ International Security, Vol. 22, No. 4 (1998),pp. 44–82. In this article, ODT means orthodox or standard ODT (defined in section 1below). In the literature, the works of Jervis, Quester, and Van Evera are usually acceptedas the foundational works of orthodox ODT.

2 See, for example, Thomas J. Christensen and Jack Snyder, ‘Chain Gangs and PassedBucks: Predicting Alliance Patterns in Multipolarity’, International Organization, Vol.44, No. 2 (1990), pp. 137–68; James D. Fearon, ‘Rationalist Explanations for War’,International Organization, Vol. 49, No. 3 (1995), pp. 401–404; Robert Gilpin, War andChanges in World Politics (Cambridge: Cambridge University Press, 1981), pp. 59–63;Charles L. Glaser, ‘Realists as Optimists: Cooperation as Self-help’, InternationalSecurity, Vol. 19, No. 3 (1994), pp. 50–90; Charles L. Glaser, ‘When Are Arms RacesDangerous?’ International Security, Vol. 28, No. 4 (2004), pp. 44–84; Ted Hopf, ‘Polarity,The Offense Defense Balance, and War’, American Political Science Review, Vol. 85, No. 2(1991), pp. 475–493; Andrew Kydd, Trust and Mistrust in International Relations(Princeton: Princeton University Press, 2005), pp. 31–33; Peter Liberman, ‘TheOffense-defense Balance, Interdependence, and War’, Security Studies, Vol. 9, No. 1 &2, 1999–2000, pp. 59–91; Evan Braden Montgomery, ‘Breaking out of the SecurityDilemma: Realism, Reassurance, and the Problem of Uncertainty’, InternationalSecurity, Vol. 31, No. 2 (2006), pp. 151–185; Barry Posen, The Sources of MilitaryDoctrines: France, Britain, and Germany between the World Wars (Ithaca: Cornell

The Chinese Journal of International Politics, Vol. 3, 2010, 213–260doi:10.1093/cjip/poq004

� The Author 2010. Published by Oxford University Press. All rights reserved.For permissions, please e-mail: [email protected]

at Fudan U

niversity on May 24, 2010

http://cjip.oxfordjournals.orgD

ownloaded from

http://cjip.oxfordjournals.org

ODT is classified in such literature as quintessential defensive realism

theory,3 and also identified as an important demarcation line between of-

fensive and defensive realism.4

The impact of ODT goes far beyond that which it has made on strategic

studies and the broader literature of realism. Its influence is such that even

non-realists such as Robert Keohane, Lisa Martin, and Alexander Wendt

have invoked ODT logic to operate arguments that counter the logic of

(offensive) realism.5

ODT has always had its critics6 who, as they become more sophisticated

over the years, often question the theory’s very foundations.7 ODT propon-

ents have responded with a vigorous defence of both its validity and utility

University Press, 1984); Barry Posen, ‘The Security Dilemma and Ethnic Conflict’,Survival, Vol. 35, No. 1 (1993), pp. 27–47; Robert Powell, ‘Absolute and Relative Gainsin International Relations Theory’, American Political Science Review, Vol. 85, No. 4(1991), pp. 1303–1320; William Rose, ‘The Security Dilemma and Ethnic Conflict: SomeNew Hypotheses’, Security Studies, Vol. 9, No. 4 (2000), pp. 1–51; and Shiping Tang‘A Systemic Theory of the Security Environment’, Journal of Strategic Studies, Vol. 27,No. 1 (2004), pp. 1–32. For more complete lists of the works that rely on ODT, see KarenRuth Adams, ‘Attack and Conquer? International Anarchy and theOffense-Defense-Deterrence Balance’, International Security, Vol. 28, No. 3 (2003/4),pp. 45–83; at pp. 45–7, footnotes 1–9 and Kier A. Lieber, War and the Engineers: ThePrimacy of Politics over Technology (Ithaca: Cornell University Press, 2005), chapter 1.

3 Jeffery W. Taliaferro, ‘Security Seeking under Anarchy: Defensive Realism Revisited’,International Security, Vol. 25, No. 3 (2000/01), pp. 128–161, at p. 135.

4 Kier A. Lieber, War and the Engineers, pp. 8–11.5 Robert O. Keohane and Lisa L. Martin, ‘The Promise of Institutionalist Theory’,

International Security, Vol. 20, No. 1 (1995), pp. 39–51, at p. 44; and Alexander Wendt,Social Theory of International Politics (Cambridge: Cambridge University Press, 1999),pp. 357–363. Stephen Van Evera, however, still complains that ODT remains underappre-ciated. Stephen Van Evera, Causes of War, p. 117. This might be true amongdecision-makers, and for good reason (see below).

6 See, for example, Colin S. Gray, Weapons Don’t Make War: Politics, Strategy, andMilitary Technology (Lawrence: University Press of Kansas, 1993), esp. chapters 1–2;Samuel P. Huntington, ‘U.S. Defense Strategy: The Strategic Innovations of the ReaganYears’, in Joseph Kruzel, ed., American Defense Annual, 1987–1988 (Lexington: LexingtonBooks, 1987), pp. 23–43, at pp. 35–7; Jack S. Levy ‘The Offensive/Defensive Balance ofMilitary Technology: A Theoretical and Historical Analysis’, International StudiesQuarterly, Vol. 28, No. 2 (1984), pp. 219–38; John Mearsheimer, ConventionalDeterrence (Ithaca: Cornell University Press, 1983), pp. 29–30; and Jonathan Shimshoni,‘Technology, Military Advantage, and World War I: A Case for MilitaryEntrepreneurship’, International Security, Vol. 15, No. 3 (1990/91), pp. 187–215. For arecent feminist whack at ODT, see Lauren Wilcox, ‘Gendering the Cult of the Offensive’,Security Studies, Vol. 18, No. 2 (2009), pp. 214–40.

7 Richard K. Betts, ‘Must War Find a Way? A Review Essay’, International Security, Vol.24, No. 2 (1999), pp. 166–198; James W. Davis, Jr., ‘Taking Offense at Offense-DefenseTheory’, International Security, Vol. 23, No. 3 (1998/99), pp. 179–182; Bernard I. Finel,‘Taking Offense at Offense-Defense Theory’, International Security, Vol. 23, No. 3, 1998/99, pp. 182–189; James D. Fearon, ‘The Offense-Defense Balance and War Since 1648’,unpublished conference paper (Stanford University, 1997), accessed Oct. 2006; Stacie E.Goddard, ‘Taking Offense at Offense-Defense Theory’, International Security, Vol. 23, No.3 (1998/99), pp. 189–195; Kier A. Lieber, ‘Grasping the Technological Peace: TheOffense-Defense Balance and International Security’, International Security, Vol. 25,No. 1 (2000), pp. 71–104 and Kier A. Lieber, War and the Engineers.

214 Tang Shiping

The Chinese Journal of International Politics, Vol. 3, 2010, 213–260

at Fudan U



ownloaded from


for understanding international politics. They tout its virtues of being struc-

tural and parsimonious,8 pointing out attempts still being made to reformu-

late, refine, and test ODT.9 As it now stands, however, the validity and

utility of ODT remain contested and unresolved.10

This article tries to advance a more definitive understanding of ODT.

I show that both ODT proponents and opponents seriously underestimate

its complexity. Specifically, the two critical components of ODT—differen-

tiation or distinguishability of weapons and military postures as either

offensive or defensive, and the offence-defence balance (ODB)—can and

should be further unpacked into several sub-dimensions.11 More critically,

these sub-dimensions have different utilities for understanding international

politics, as some are valid, operable and useful, and some are not.

I hence unpack into sub-dimensions these two major components of ODT

and make critical examinations of each. To prevent any further waste of

intellectual resources on futile research, I criticize those that are invalid,

inoperable and useless and at the same time reformulate and rebuild those

that are valid, operable, and useful, thus providing directions towards more

fruitful research. I show that having spent relatively more resources on in-

valid ODT components, we have yet fully to appreciate the implications of

theories that are valid, and that deepening our understanding of these com-

ponents will yield concrete theoretical and policy payoffs. Although my

endeavour is necessarily a dirty job that is bound to ruffle many feathers,

it nevertheless contributes to IR literature, especially security studies, as well

as to certain contemporary policy debates.

8 Charles L. Glaser and Chaim Kaufmann, ‘Correspondence: Taking Offense atOffense-Defense Theory’, International Security, Vol. 23, No. 3 (1998/99), pp. 200–6;Sean M. Lynn-Jones, ‘Offense-Defense Theory and Its Critics’, Security Studies, Vol. 4,No. 4 (1995), pp. 672–4 and Stephen Van Evera ‘Correspondence: Taking Offense atOffense-defense Theory’, International Security, Vol. 23, No. 3 (1998/99), pp. 195–200.Stephen Van Evera claims that ODT is a ‘master theory’ and ‘the most powerful and usefulRealist theory on the causes of war’. Stephen Van Evera, Causes of War, p. 117; emphasisadded.

9 Karen Adams, ‘Attack and Conquer?’; Stephen Biddle, ‘Rebuilding the Foundation ofOffense-Defense Theory’, Journal of Politics, Vol. 63, No. 3 (2001), pp. 741–74; YoavGortzak, Yoram Z. Hazfel, and Kevin Sweeney, ‘Offense-Defense Theory: AnEmpirical Assessment’, Journal of Conflict Resolution, Vol. 49, No. 1 (2005), pp. 67–89;Kier A. Lieber, ‘Grasping the Technological Peace: The Offense-Defense Balance andInternational Security’, International Security, Vol. 25, No. 1 (2000), pp. 71–104; andKier A. Lieber, War and the Engineers. Biddle’s work is both a critique of orthodoxODB and an attempt to offer an unorthodox formulation of ODB. I do not deal indepth here with unorthodox ODB.

10 For a summary of more recent debate on ODT, see Sean Lynn-Jones ‘DoesOffense-Defense Theory Have a Future?’ http://www.ciaonet.org/wps/lys03/lys03.pdf(accessed on July 4, 2008).

11 The other critical component is geography. Ignoring it, however, does not jeopardize thearguments below.

Offence-defence Theory 215


at Fudan U



ownloaded from


My discussion leads to four general conclusions. First, that objective ODB

is essentially a theoretical hoax that exists only in the imagination of ODB

proponents and which hence merits no further intellectual resources.

Second, although subjective ODB might be useful for understanding inter-

national politics, much of existing research on subjective ODB has yet to

touch upon the most fundamental aspects of it. Third, although differenti-

ation of military postures is possible, that for most weapons, upon which

existing discussion unfortunately focuses, is not. Existing discussion on dif-

ferentiation has also missed the real difference between offensive realism on

the one side and non-offensive realism theories on the other side. As a result,

theories on signalling benign intentions through military postures—which

obviously hinge on military posture differentiation—remain underdevel-

oped.12 Fourth, classifying ODT as a quintessential defensive realism

theory or identifying it as a demarcation line between offensive realism

and defensive realism is at best simplistic and at worst incorrect.

Before proceeding, I specify three caveats. First, I am fully aware that

others have raised the argument that ODT literature consists of a group of

arguments rather than a body of theories.13 I disagree. Although a portion

of ODT literature indeed comprises a raft of various statements, there is

much that also presents theories or proto-theories, if theory is defined as a

scientific explanation of a phenomenon or phenomena. Moreover, most

ODT proponents subscribe to two common key propositions (see the first

section). As such, ODT should be treated as a body of related theories or

proto-theories, although not a single theory.

Second, over the years the logic of ODT has been employed for under-

standing many issues, such as causes of war, problems of cooperation and

ethnic conflict. During discussion I touch upon certain of these applications

whenever appropriate, but rather than dealing with them extensively or

in-depth, my focus is on the fundaments of ODT. The rationale here is

that a better grounding in this theory makes apparent that certain ODT

applications are misguided and should be rejected or reformulated, and

certain others are underdeveloped and should be strengthened.

Third, because most proponents of ODT deal exclusively with the

pre-mutual assured destruction (MAD) era or with human history itself

(which more or less covers the pre-MAD era) I first deal with ODT as if

it were purely a theory of war and peace in the pre-MAD era. I return to the

relationship between nuclear weapons and ODT only after considering the

validity and utility of ODT in the pre-MAD or conventional era.

12 For earlier discussions, see Charles Glaser, ‘Realists ad Optimists’; Andrew Kydd, Trustand Mistrust in International Relations; and Evan Montgomery, ‘Breaking out of theSecurity Dilemma’.

13 Karen Adams, ‘Attack and Conquer?’ p. 46, footnote 5.

216 Tang Shiping


at Fudan U



ownloaded from


The rest of this article is in seven sections. First section defines and delin-

eates ODT and clarifies the miscellaneous, different levels of analysis applic-

able to discussing a state’s military and national military strategy. The

section ‘Differentiation of Military Postures and Weapons’ examines the

possible differentiation of military postures and weapons into offensive

and defensive types. The section ‘The Objective ODB’ addresses the object-

ive interpretations of ODB and section ‘The Subjective ODB’ addressed the

subjective interpretations of ODB. The section ‘Defending and Testing the

Balance’ investigates some of the more systematic defences so far of ODB

and also recent empirical tests of ODB and war to reveal that ODB is

essentially indefensible and that empirical evidence for ODB is at best

weak. The section ‘Does ODB have a MAD/Nuclear Future?’ addresses

the validity and utility of ODB in the MAD era. The section

‘Implications’ rigorously re-classifies ODT and pinpoints certain implica-

tions for future work. A brief conclusion follows.

Definitions and Clarifications

Defining ODT

Perhaps surprisingly, even after three decades of fierce debate and many

articles, neither proponents nor opponents of ODT have explicitly defined

the theory. Most have never bothered; many have taken it as essen-

tially equivalent to or centred upon ODB.14 Lynn-Jones and Glaser and

Kaufmann meanwhile assert that measuring ODB requires no differenti-

ation of weapons.15 As will later become clear, however, although measuring

ODB does not necessarily depend on the differentiation of weapons, it does

depend on differentiation at levels higher than that of weapons, such as

military postures. Discussions on ODB inevitably involve differentiation

at certain levels.

A brief clarification of these areas16 makes clear that (orthodox/standard)

ODT consists of a body of theories that operate upon two principal prop-

ositions (and variables).17 First, differentiation of weapons and military

postures is not only possible but useful (e.g. for understanding the regulation

14 See for example, Karen Adams, ‘Attack and Conquer’; Yoav Gortzak et al.,‘Offense-Defense Theory’, pp. 67–9; Kier Lieber, ‘Grasping the Technological Peace’,pp. 71–2; Kier Lieber War and the Engineers, pp. 1–2; Stephen Biddle, ‘Rebuilding theFoundation’, pp. 744–6; and Stephen Van Evera, Causes of War, 117, footnote 1.

15 Sean Lynn-Jones, ‘Offense-Defense Theory’; 672–7; Sean Lynn-Jones, ‘DoesOffense-Defense Theory Have a Future?’ p. 8; Charles Glaser and Chaim Kaufmann,‘What is Offense-Defense Balance and How Can We Measure It?’ pp. 79–80.

16 Another divergent point is whether ODT is equivalent to the security dilemma theory. Iaddress this point in section ‘Implications’.

17 Glaser and Kaufmann, and Lynn-Jones came close to the definition advanced here. See,Charles Glaser and Chaim Kaufmann, ‘What Is the Offense-Defense Balance and HowCan We Measure It?’ pp. 47–48; Sean Lynn-Jones, ‘Offense-Defense Theory’, pp. 665;Sean Lynn-Jones, ‘Does Offense-Defense Theory Have a Future?’ p. 15, footnote 31. The



at Fudan U



ownloaded from


of the security dilemma and designing arms control schemes). Second, war

and conquest are more likely when offence has, or is perceived to have an

objective advantage, and vice versa—that is to say, ODB influences the

outbreak of war.

Levels of Analysis: Military Power and Military Strategy

Discussions of state military power and (military) strategies within ODT

literature and that of broader strategic studies employ several (eight to be

exact) different concepts—or more precisely, levels of analysis.18 Even

though each of these concepts might have different meanings for different

scholars, authors of these discussions nevertheless neglect to make rigorous

definitions of them or to delineate the relationships among them.

It is useful to divide these concepts into two general categories: concepts

that denote the physical dimension and those that denote the strategic di-

mension of a state’s military (see Table 1). From the lowest to the highest

level, concepts that denote the physical dimension include weapon, arsenal,

military capability, and total war-fighting power. From the lowest to the

highest level, concepts that denote the strategic dimension include military

posture, military doctrine, military strategy, and (national) grand strategy.

As a principle, concepts at the higher level subsume or drive concepts at a

lower level. Total power hence subsumes military capability, which in turn

subsumes arsenals and weapons. Similarly, grand strategy subsumes and

drives military strategy, which in turn both subsumes and drives military

doctrine and military posture.19 In the strategic dimension, because thinking

at the higher level drives thinking at a lower level, that at lower level, for

Table 1 Levels of Analysis: Military Power, Strategy, and Outcomes

Physical Components Strategic Components Outcomes

Weapons Military posture/stance Battle

Arsenals Military doctrine Operation

Military capabilities Military strategy Campaign

Total war-fighting power Grand strategy/security strategy War

foundation of these two propositions, of course, came from Robert Jervis, ‘Cooperationunder the Security Dilemma’, pp. 187–214.

18 I understand that ‘level of analysis’ in IR is often linked with Waltz’s scheme. I retain thislabel because I have not been able to find one more suitable. Plus, Waltz does not own‘level of analysis’. See, Kenneth A. Waltz, Man, the State, and War (New York: ColumbiaUniversity Press, 1959).

19 Douglas Porch, ‘Military ‘Culture’ and the Fall of France in 1940’, pp. 168–9; BarryPosen, Sources of Military Doctrine, p. 7 and Jonathan Shimshoni, ‘Technology,Military Advantage, and World War I’, pp. 187–8. Huntington develops a scheme thatincludes only four concepts (two each for each dimension): weapons and weapon technol-ogies, military capabilities, military strategy, and political goals His ‘political goals’ is

218 Tang Shiping


at Fudan U



ownloaded from


example posture, can also be understood as a partial indicator of thinking at

the higher level, for example, doctrine.20

Weapons are simply the equipment possessed by a state’s military, and

nothing more. If there is a pure technological component within a state’s

military, the level of weapon is it: Anything above the level of weapon can

no longer be purely technological.

A state’s arsenal is the total sum of the weapons, rather than a single

weapon or random arrays of weapons, that a state deploys. Weapons in

an arsenal have been chosen and combined to achieve specific military ob-

jectives on the battlefield. Because the choice of weapons within an arsenal

and their combination to achieve different military objectives is mostly de-

pendent on human decisions, an arsenal is no longer a purely technological

outcome or phenomenon. Military capability is simply the total power of

a state’s military, or ‘the overall size, organization, training, equipment,

logistic support, and the leadership of a military force’.21 A state’s total

war-fighting power is its latent war-fighting capability, which can be roughly

measured according to a state’s total gross national product (GNP).22

Military posture includes operational doctrines (e.g. operational tactics

and operational rules of engagement), patterns of troop deployment (i.e. for-

ward or non-forward deployment), and deployment of weapons.23

Military doctrine is ‘the way an army organizes to fight, that is, the pro-

cedures and methods it applies in combat’.24 In general, there are two ideal

types of military doctrine. They are: offensive doctrine and defensive doc-

trine, the latter of which includes deterrent doctrine.25

Military strategy, also sometimes called military policy, comprises a state’s

military strategic goals and the means to achieving them.26 The US ‘Fighting

roughly equivalent to grand strategy here. Samuel Huntington, ‘U.S. Defense Strategy’,pp. 35–7.

20 I thank Taylor Fravel for this formulation.21 Samuel Huntington, ‘U.S. Defense Strategy’, p. 36. Obviously, a state’s military capability

is not a purely technological outcome (see below).22 John Mearsheimer, The Tragedy of Great Power Politics (New York: Norton, 2001), chap-

ter 4. For realists, power is mostly material and it is material power that ultimately decidesoutcomes in international politics. But see section ‘The Objective ODB’.

23 Douglas Porch, ‘Military ‘Culture’ and the Fall of France in 1940: A Review Essay’,International Security, Vol. 24, No. 4 (2000), pp. 157–80, at p. 179. Jervis’s militarystance is close to military posture here. Robert Jervis, ‘Cooperation under the SecurityDilemma’, p. 199.

24 Douglas Porch, ‘Military ‘Culture’ and the Fall of France in 1940’, p. 168. Porch, however,denies that doctrine per se can be differentiated into offensive or defensive. My definitionof military doctrine is close to the examples of military doctrines given by Posen. See BarryPosen, Sources of Military Doctrine, pp. 14–15.

25 Glenn H. Snyder, Deterrence and Defense: Toward a Theory of National Security(Westport: Greenwood Press, 1961). See also Barry Posen, Sources of Military Doctrine,pp. 7, 14–15.

26 Charles L. Glaser, ‘Political Consequences of Military Strategy: Expanding and Refiningthe Spiral and Deterrence Models’, World Politics, Vol. 44, No. 2 (1992), pp. 497–538.Glaser seems to take military policy to mean both military posture and military strategy.



at Fudan U



ownloaded from


and prevailing in two regional wars’ and China’s ‘winning local conflicts

under hi-tech circumstances’ are examples of states’ military strategy.

The concept of grand strategy is perhaps the least ambiguous in the lit-

erature under scrutiny. Narrowly defined, grand strategy is a state’s security

strategy and mostly about military and diplomatic (e.g. alliances) means

toward states’ security goals. A broader definition of grand strategy is

that of the total sum of a state’s political, economic, diplomatic and military

strategies.27

Because part of ODT (especially objective ODB) explains the outcomes of

actual military conflicts, it is also necessary to differentiate four levels of

analysis of military outcomes. Again from lowest to the highest, they are:

operation, battle, campaign, and war.28

This clarification is extremely important because many, either uninten-

tionally or otherwise, have conflated various different levels of analysis to

support their theses.For example, when arguing against any forms of arms

control, Colin Gray asserts, ‘Rarely can war be won by defensive strategies

alone’.29 Such a statement is obviously valid only if ‘defensive strategies’

means defensive military tactics (i.e. not taking initiatives when trying to

defeat the aggressor). When ‘defensive strategies’ means defensive grand

strategies or even defensive military strategies, then a (defensive) war can

be won by defensive grand strategies with a defensive military strategy but

under an offensive military doctrine.30

Elizabeth Kier rejects realism’s explanations for the origins of military

doctrines and advances a supposedly superior cultural explanation of why

Britain and France were unable to adopt an offensive military strategy and

grand strategy when Germany did, both post WWI and before WWII. She

fails, however, to differentiate the concepts of (grand) strategy and military

doctrine and therefore to acknowledge that it is strategy that determines

military doctrine rather than the other way around.31 Kier fails to grasp

that because France and Britain were more benign states, they both ration-

ally adopted a defensive grand strategy and a more defensive military doc-

trine. Hitler’s Germany, in contrast, as a state bent on expansionism,

adopted an offensive grand strategy and an offensive military doctrine

27 See, for example, Barry Posen, Sources of Military Doctrines, p. 13; Douglas Porch,‘Military Culture and the Fall of France in 1940’, p. 168. Robert J. Art, ‘A DefensibleDefense: America’s Grand Strategy after the Cold War’, International Security, Vol. 15,No. 4 (1991), pp. 5–53, at pp. 6–7.

28 Stephen Biddle, ‘Rebuilding the Foundation of Offense-Defense Theory’, p. 747, footnote8. Glaser and Kaufmann list three levels: strategic, operational, and tactical. See, CharlesGlaser and Chaim Kaufmann, ‘What is Offense-Defense Balance’, pp. 54–5. I believe thatBiddle’s scheme is more fine-grained and sound.

29 Colin Gray, Weapons Don’t Make War, p. 28; emphasis added.30 Samuel Huntington, ‘U.S. Defense Strategy’, p. 37.31 Elizabeth Kier, Imagining War: French and British Military Doctrines between the Wars

(Princeton: Princeton University Press, 1996).

220 Tang Shiping


at Fudan U



ownloaded from


simply because defensive strategies and doctrines cannot conquer.32

Realism’s explanation is hence far more straightforward and convincing

than Kier’s supposedly superior cultural explanation.

Posen takes military doctrine to include at three aspects: ‘its offensive,

defensive, or deterrent character; its coordination with foreign policy;

and the degree of innovation it contains.’ So defined, Posen’s military doc-

trine is far broader than that most authors would agree upon; his definition

of military doctrine is closer to military strategy or even grand strategy. But

the examples of military doctrines given by Posen are far narrower—they are

mostly concerned with the offensive, defensive, or deterrent nature of a

state’s military doctrine, without any foreign policy input.33

Finally, Jonathan Shimshoni fails to appreciate that Jack Snyder’s call for

NATO to limit offensive conventional forces in the European theatre during

the Cold War was actually a call for confidence-building measures at the

posture or even doctrine-level and intended to signal moderation and reduce

tension between the two opposing camps.34 Snyder’s call was thus a genuine

measure of reassurance (see below) and not an ill-advised attempt towards

technological fixes for the Cold War.

Differentiation of Military Postures and Weapons

Differentiation of states’ aspects into offensive or defensive types is one of

the cornerstones of ODT; without it, the whole ODT enterprise is on shaky

ground.

Differentiation: Logic and Utility

ODT proponents have so far been neither explicit nor consistent on (i)

whether the focus of differentiation should be on the physical or strategic

components of a state’s military, and (ii) whether differentiation should be

absolute or relative. Because not resolving these two crucial issues has

caused much confusion, this section sets out to resolve them and to under-

score the actual value of differentiation.

Weapons occupy the lowest level in the physical component of a state’s

military, and military postures the lowest level in its strategic component.

I first examine the possibility of differentiation at these two levels.

Much of the existing discussion on differentiation has been inexplicit

about whether the focus should be on weapons or military postures,

32 Douglas Porch, ‘Military Culture and the Fall of France in 1940’, pp. 168–9.33 Barry Posen, Sources of Military Doctrine, p. 7. Posen provides several examples in

pp. 14–15.34 Jack Snyder, ‘Limiting Offensive Conventional Forces: Soviet Proposals and Western

Options’, International Security, Vol. 12, No. 4 (1988), pp. 48–77; Jonathan Shimshoni,‘Technology, Military Advantage, and World War I’, p. 188, footnote 3.



at Fudan U



ownloaded from


or generally on weapons. I propose that the correct focus is a state’s military

postures, for two reasons.

First, although certain weapons can indeed be classified as purely offen-

sive or purely defensive when viewed in isolation,35 most weapons have dual

uses. Hence, ‘whether a weapon is offensive or defensive often depends on

the particular situation . . . the way in which the weapon is used’.36 Although

Lynn-Jones appears to be defending the differentiation of weapons, he does

not inform us how weapons might consistently be differentiated as either

offensive or defensive.37

Second, states deploy arsenals that comprise weapons and weapon sys-

tems rather than a random array of weaponry, and generally employ weap-

ons in combination rather than in isolation. This means that (i) offence and

defence generally depend upon one another to be effective, and (ii) offensive

weapons can be deployed both for defensive support purposes and vice

versa.38 For instance, ‘anti-aircraft weapons seem obviously defensive.

. . .But the Egyptian attack on Israel in 1973 would have been impossible

without effective air defence that covered the battlefield.’39 To take this

point still further, even fortification serves this dual purpose by freeing

forces which can then be deployed for offensive purposes.40

Consequently, differentiation of weapons into offensive or defensive types

is generally difficult or meaningless, even if it is possible.41 Unsurprisingly,

proponents of weapons differentiation themselves mostly talk about matters

above the level of weapons or technology. Although when discussing

differentiation Jervis puts weapons ahead of (military and political) policies,

and is inexplicit about which takes precedence, his discussion nevertheless

focuses mostly on military postures (or stances) and policies.42 Goldfischer’s

35 For example, viewed in isolation, fortification and surface-to-air missiles (SAMs) arealmost purely defensive whereas aircraft carriers are almost purely offensive.

36 Robert Jervis, ‘Cooperation under the Security Dilemma’, p. 202; Jack Levy, ‘TheOffensive/Defensive Balance of Military Technology’, pp. 225–7; Richard Betts, ‘MustWar Find a Way?’ pp. 185–186; and Samuel Huntington, ‘U.S. Defense Strategy’, p. 36.

37 Sean Lynn-Jones, ‘Offense-Defense Theory and Its Critics’, pp. 674–7.38 Jack Levy, ‘Offensive/Defensive Balance of Military Technology’, pp. 225–7. States of

course deploy different arsenals for different military operations. An arsenal for an offen-sive campaign and strategies will necessarily be different from an arsenal for defensivecampaign and strategies. But this does not mean that individual weapons can bedifferentiated.

39 Robert Jervis, ‘Cooperation under the Security Dilemma’, pp. 202–3.40 George Quester, Offense and Defense in the International System, p. 63.41 The fundamental cause behind this difficulty of differentiating weapons has been evolu-

tion. Before our ancestors came to confront each other, they had to confront many power-ful non-human predators. As such, our ancestors must accumulate some capabilities ofkilling even if they merely want to defend in order to survive. Thus, weapons, from its verybeginning as tools of the early humans, have always been of the dual purposes of killingand avoiding being killed. Human beings’ carnivorous nature adds another impetus fordeveloping the skill to kill. Robert O’Connell, Of Arms and Man: A History of War,Weapons, and Aggression (Oxford: Oxford University Press, 1989), pp. 20–2.

42 Robert Jervis, ‘Cooperation under the Security Dilemma’, pp. 199–206.

222 Tang Shiping


at Fudan U



ownloaded from


discussion in defence of differentiating weapons is also mostly about mili-

tary capabilities.43 In a discussion in the same vein Van Evera, too, talks

about ‘modern guerrilla war’ rather than weapons.44 Similarly, Glaser

consistently focuses on military strategies and policies rather than on weap-

ons.45 Finally, even George Quester in his foundational work on ODT dis-

cusses mostly military capabilities rather than weapons.46

Because most weapons in state arsenals can be employed for both offen-

sive and defensive reasons, even a purely defensive realist state will neces-

sarily deploy offensive weapons and capabilities.47 Consequently, apart from

the fact that differentiation of all the weapons in any arsenal is in any event

unrealistic, one cannot identify a state that deploys an offensive weapon—

even when that weapon is unequivocally so—as one with offensive inten-

tions. On the other hand, however, although a state’s military posture is the

lowest denominator in the strategic component of a state’s military, it at

least partly reflects the state’s political thinking behind its potential use of

force and hence its intentions. As a result, differentiating a state’s military

posture into either offensive or defensive (relatively speaking, of course)

provides a glimpse into its intentions.48

Finally, it should be said that differentiation of military postures can only

be relative, simply because all militaries have both offensive and defensive

capabilities. Absolute differentiation is impossible other than in extreme

situations (e.g. when a state’s arsenal contains only tanks). In other

words, we can say that one state’s military posture is more offensive than

43 David Goldfischer, The Best Defense: Policy Alternatives for US Nuclear Security from the1950s to the 1990s (Ithaca: Cornell University Press, 1993), chapter 2.

44 Stephen Van Evera ‘Correspondence: Taking Offense at Offense-defense Theory’,pp. 195–6.

45 Charles Glaser, ‘Political Consequences of Military Strategy’; Charles Glaser, ‘Realists asOptimists’; and Charles Glaser, ‘When Are Arms Races Dangerous?’

46 George Quester, Offense and Defense. Quester reiterates this notion in his new introductionto the new edition of his book. George Quester, ‘Introduction’, in Offense and Defense inInternational System, pp. x–xv.

47 Stephen Biddle, ‘Rebuilding the Foundation of Offense-Defense Theory’, p. 769; andRobert Jervis ‘Cooperation under the Security Dilemma’, pp. 201–3.

48 Even if states state their military strategy, security strategy, and grand strategies in whitepapers, they never say that their strategies are offensive or aggressive. Although differen-tiation can be achieved at all four levels of the strategic component, the higher the level ofa component, the more secretive it tends to be. By comparison, military postures–or atleast certain components of a state’s military posture (e.g. forward or non-forward troopdeployment) can be easily observed or more readily spied upon than say, grand strategiesand military strategies. Barry Posen, Source of Military Doctrines, p. 16. When the ultim-ate utility of differentiation is for gauging another state’s intention or differentiating amalign state from a benign state (see below), military postures should be the correct focusof differentiation. For example, based on open sources, M. Taylor Fravel is able to con-clude that China’s military posture is largely defensive. Taylor Fravel, ‘Securing Borders:China’s Doctrine and Force Structure for Frontier Defense’, Journal of Strategic Studies,Vol. 30, No. 4/5 (2007), pp. 705–37.



at Fudan U



ownloaded from


that of another state, but we cannot call one state’s military posture offen-

sive without a reference point.49

The False Battle over Differentiation

In one of the most systematic defences of ODT, Lynn-Jones correctly points

out that although repeatedly denying that weapons can be differentiated,

almost all critics of ODT regularly employ the differentiation of military

capabilities, military postures, military strategies, and grand strategies to

advance their arguments.50

For instance, although explicitly denying that conventional weapons

can be meaningfully differentiated as either offensive or defensive,51

Mearsheimer has written a book on conventional deterrence that is never-

theless essentially about how, by adopting certain military postures, one state

can deter another from taking offence. Similarly, although Colin Gray

also explicitly denies that weapons—including nuclear weapons which are

generally understood as the ultimate defensive weapon—can be differen-

tiated, he nevertheless repeatedly talks about offensive and defensive policies

and strategies.52

Finally, Samuel Huntington is most unequivocal. ‘The offence/defence

distinction is somewhat more useful when it comes to talking about military

capabilities. Here the reference is to the overall size, organization, training,

logistic support, and the leadership of a military force. Depending upon how

these various elements are combined, some military forces will be better

prepared to fight offensive actions, while others will be better prepared to

fight defensive actions . . .The distinction between offence and defence

is. . .applicable to how military can be used—that is to say, to strategy. . .

The offence/defence distinction is also relevant at a higher level of analysis

beyond strategy, which is concerned with the overall foreign policy goals of

a state and that state’s willingness to initiate the use of military force

to achieve those goals. . . . [Thus,] useful distinction can be drawn between

offensive and defensive policy goals, strategies, and capabilities.’ But

Huntington is also unequivocally against the differentiation of weapons:

‘weapons may be usefully differentiated in a variety of ways, but the

offence/defence distinction is not one of them.’53

49 The fact that military postures can only be differentiated relatively also means that dif-ferentiation is an essentially dyadic variable.

50 Sean Lynn-Jones, ‘Offense-Defense Theory and Its Critics’, pp. 672–6, footnotes 27, 35, 41and 44.

51 John Mearsheimer, Conventional Deterrence, pp. 25–7; and John Mearsheimer, LiddellHart and the Weight of History (Ithaca: Cornell University Press, 1988), p. 36, 44, and61; John Mearsheimer, ‘Back to the Future: Instability in Europe after the Cold War’,International Security, Vol. 15, No. 1 (1990), pp. 5–56, at p. 13, footnote 14. Mearsheimerexplicitly states that differentiation is only relevant at the nuclear level (see below).

52 Colin Gray, Weapons Don’t Make War.53 Samuel Huntington, ‘U.S. Defense Strategy’, pp. 36–7; emphasis added.

224 Tang Shiping


at Fudan U



ownloaded from


In sum, opponents of ODT have never seriously questioned the fact that

military capabilities, military postures, military strategies, and grand strate-

gies can be differentiated into (relatively) offensive or defensive types. It is

the differentiation of weapons with which they take issue. Opponents of

ODT refuse to accept that, other than perhaps a few, weapons can be

differentiated into offensive or defensive types.54

Meanwhile, even proponents of ODT often admit that differentiation

of weapons is problematic, if not extremely difficult. For instance, Jervis

admits, ‘[No] simple and unambiguous definition [of offensive or defensive

weapons] is possible and in many cases no judgment can be reached.’55 If

this is indeed so, opponents and proponents of ODT actually differ little on

this particular issue of differentiation of weapons.56

Certain proponents of ODT have unfortunately either failed to notice the

fact that there is actually little disagreement between them and their oppon-

ents on the matter of differentiation of military capabilities, military pos-

tures, military strategies, and grand strategies. As a result, they have spent

much energy in defending the essentially invalid differentiation of weapons.

Goldfischer offers perhaps the most spirited defence of the possibility of

offensive and defensive weapon differentiation. Referring to Huntington’s

criticism of differentiating of weapons,57 Goldfischer charges that

Huntington ignores weapons: ‘Huntington’s definition of capability excludes

the contribution of particular types of weapons. (‘Equipment’ seems at

best an unnecessary indirect reference to missiles, bombers, tanks, or

anti-ballistic missile systems.) That omission allows him to suggest that

(presumably as a representative example) the switch from a defensive to

an offensive capability can be achieved merely on providing gasoline.58

But Huntington does not deny the contribution of equipment or weapons

to military capabilities, and certainly does not suggest that switching from

defence to offence can be achieved simply by providing gasoline. Moreover,

Huntington is explicit in his assertion that military capabilities (of which

weapons are only a part) and other matters on a level higher than military

capabilities (e.g. strategies) can be differentiated. Goldfischer thus charges

Huntington of a crime he never committed.

The Real Battle over Differentiation

Many proponents of ODT fail to recognize the real difference between them

and their opponents on the differentiation issue, for two reasons. First,

54 Kier Lieber, War and the Engineers, pp. 34–44.55 Robert Jervis, ‘Cooperation under the Security Dilemma’, p. 201.56 Defensive realists, of course, have long argued that military doctrines and other strategic

components of a state’s military can be differentiated into offensive and defensive types(e.g. Charles Glaser 1992; Robert Jervis 1978; Barry Posen 1984, pp. 13–15).

57 Samuel Huntington, ‘U.S. Defense Strategy’, pp. 35–7.58 David Goldfischer, The Best Defense, p. 31.



at Fudan U



ownloaded from


certain proponents of ODT have emphasized two invalid utilities of differ-

entiation. One is the measurement of (objective) ODB which, in reality

cannot be measured even if differentiation of weapons is possible (see section

‘The Objective ODB’).59 Second, they have emphasized is the differentiation

of weapons which, in reality, cannot be meaningfully differentiated.

The valid utility of differentiation is to serve as part of the foundation for

constructing a theory of signalling intentions through military postures (not

weapons), especially a theory of signalling benign intentions through defen-

sive military postures. A theory of signalling benign intentions through

defensive military postures—which forms an integral part of the defensive

realism theory of cooperation-building—critically depends on differentiating

military postures into offensive and defensive types.60 When military pos-

tures cannot be differentiated, signalling benign intentions through military

postures as advocated by defensive realism becomes very difficult, although

not impossible. In contrast, when military postures can be differentiated,

signalling benign intentions through military postures as advocated by de-

fensive realism becomes less formidable, although still both costly and

risky.61

Offensive Realism against Differentiation: Unnecessary Offence

Once we recognize the valid level and utility of differentiation, the real dif-

ference between certain proponents of ODT (mostly defensive realists) and

opponents of ODT (mostly offensive realists) becomes evident.

Because the whole logic of offensive realism will collapse if cooperation,

other than a temporary alliance when facing a common threat, is a viable

means of external self-help under anarchy,62 offensive realists have been

trying hard to deny that cooperation is a viable means of external self-help

under anarchy.63 And because defensive realism theory’s of cooperation-

building partly depends on signaling benign intention with defensive military

postures and in turn the possibility of differentiation weapons or postures,

59 Certain proponents of ODB have argued that differentiation is not necessary for measur-ing ODT. For my discussion on this point, see footnote 13 above.

60 I develop reassurance as a defensive realism theory of cooperation-building in ShipingTang, A Theory of Security Strategy for Our Time: Defensive Realism (New York: PalgraveMacmillan 2010), chap. 5.

61 Signals, whether for resolve or for benign intentions, must be somewhat costly and carrysome risk in order to convey true intentions or be considered as credible. ThomasSchelling, Arms and Influence (New Haven: Yale University Press, 1966); James Fearon,‘Signaling versus the Balance of Power and Interest’, Journal of Conflict Resolution, Vol.38, No. 2 (1994), pp. 236–69; Andrew Kydd, Trust and Mistrust in International Relations;Shiping Tang, A Theory of Security Strategy, Chapter 5.

62 Shiping Tang, ‘Fear in International Politics: Two Positions’, International Studies Review,Vol. 10, No. 3 (2008), 451–70.

63 John Mearsheimer, Tragedy of Great Power Politics, pp. 32–6.

226 Tang Shiping


at Fudan U



ownloaded from


offensive realists have sought to undermine the logic of differentiation, thus

in turn the possibility of achieving cooperation under anarchy.64

Consequently, what offensive realists actually want to deny as regards the

problem of differentiating weapons or military postures is not that differen-

tiation is in itself possible, but rather that cooperation under anarchy is

actually possible. To achieve this, in addition to denying that there have

ever been cases of successful cooperation-building through reassurance and

even of signalling benign intentions,65 offensive realists also wage a

two-front assault against the defensive realism logic of cooperation under

anarchy.

First, because defensive realists believe that states can achieve cooperation

through costly signalling of benign intentions, which partly depends on the

differentiation of military postures and weapons,66 offensive realists argue

that such costly signalling of benign intention is difficult to initiate and

almost impossible to achieve, even if military postures and weapons actually

can be differentiated.67 This offensive realism stance, however, is logically

untenable and empirically false.

Logically speaking, signalling benign intentions is possible whether or not

military postures cannot be differentiated. A state can, in principle, signal

benign intentions simply by reducing the arms and troops along its border.

The problem, of course, is that such a move is so risky that no states would

ever try it. Because military postures can always be differentiated relatively,

however, states can, if they choose, always signal benign intentions within

some acceptable level of risk.68

64 While many offensive realists have been less than explicit in their motives for denying thepossibility of differentiation (David Goldfischer 1993, pp. 16, 22, 26–32), at least one ofthem—John Mearsheimer—has been explicit on this point. John J. Mearsheimer,‘Interview’, International Relations, Vol. 20, No. 1 (2006), pp. 123, 231–4. Interestingly,Colin Gray, another offensive realist and a prominent critic of ODT, has not only expli-citly argued that military postures and policies can be differentiated, but that defensivepostures and policies can reassure other states (i.e. ‘dampen foreign anxieties’) and reducethe chance of inadvertent war (Colin Gray, Weapons Don’t Make War (1993), p. 22).

65 Interestingly, some scholars who may be closer to defensive realism engage in this activity.See, for example, David M. Edelstein, ‘Managing Uncertainty: Beliefs about Intentionsand the Rise of Great Powers’, Security Studies, Vol. 12, No. 1 (2002), pp. 38–40; KierLieber, War and the Engineers, p. 5; and Evan Montgomery, ‘Breaking out of the SecurityDilemma’, p. 153.

66 Robert Jervis, ‘Cooperation under the Security Dilemma’; Charles Glaser, ‘Realists asOptimists’; Evan Montgomery, ‘Breaking out of the Security Dilemma’ and AndrewKydd, Trust and Mistrust in International Relations, p. 183.

67 John Mearsheimer, ‘Interview’, pp. 123; 231–4. Offensive realists have also waged anotherassault on the possibility of cooperation under anarchy, arguing that cooperation is dif-ficult to achieve and sustain because of states’ concern about relative gains from cooper-ation. I skip this issue because it is not directly relevant to the discussion here. Moreover,this is a false issue, as Randall Schweller points out. Randall Schweller, ‘Neorealism’sStatus Quo Bias: What Security Dilemma?’ Security Studies, Vol. 5, No. 3 (1996),pp. 91–121, at pp. 109–10.

68 Even when military postures can be differentiated, signalling benign intention involves realrisk. In fact, a signal of benign intention will not carry any credibility unless it incurs



at Fudan U



ownloaded from


Empirically speaking, certain statesmen have not only employed the

signalling of benign intention but succeeded in obtaining cooperation.

For instance, the 1904 entente cordiale between Britain and France contrib-

uted to the détente between them. Likewise, a series of military

confidence-building measures along their long border facilitated a strategic

partnership between post-Soviet Russia and China.69

Second, because certain defensive realists suggest that signalling benign

intentions towards cooperation depends on the differentiation of weapons

(and the ODB, see below), offensive realists strive to undermine the logic of

signalling benign intention by denying that differentiation of weapons is

possible, and by attacking the general validity of ODT (which centres on

differentiation and ODB). On this point, the offensive realist stand is cer-

tainly valid; weapons often cannot be differentiated.

That weapons generally cannot be differentiated, however, does not

undermine the defensive realist logic of signalling benign intentions towards

cooperation, because such signalling ultimately depends on military postures

(or other strategic military components above that of military posture) and

not on weapons, as many defensive realists mistakenly maintain. Moreover,

there are means of signalling benign intentions other than that of military

postures.70

In reality, the offensive realism stand against differentiation (in order to

undermine the possibility of signalling benign intentions) is unnecessary.

When a state believes that other states are inherently aggressive—whether

by nature (as ‘human nature [offensive] realism’ holds it) or compelled by

anarchy (as ‘structural offensive realism’ holds it)—and other states are

certain to take advantage of your good will, cooperation—barring tempor-

ary alliance when facing a common threat—becomes inherently irrational.71

As such, there is no rationale for seeking cooperation and thus none for

signalling benign intentions (through military postures or not), even if both

differentiation of postures (or weapons) and signalling benign intentions is

possible. The offensive realism assumption that states are aggressive or must

be aggressive obviates any need for further arguing that cooperation is risky,

because cooperation has already been made logically impossible. Offensive

certain costs to the sender of the signal. For details, see Shiping Tang, A Theory of SecurityStrategy, Chapter 5.

69 Certain structural realists have thus chosen to ignore these cases of successful cooperation.See, for example, David Edelstein, ‘Managing Uncertainty’, pp. 38–40; and EvanMontgomery, ‘Breaking out of the Security Dilemma’, p. 153. The fact that these casesof cooperation were facilitated by a common threat does not invalidate the notion thatcostly signalling contributes to cooperation.

70 I elaborate on these measures in Shiping Tang, A Theory of Security Strategy, Chapter 5.71 Eric J. Labs, ‘Beyond Victory: Offensive Realism and the Expansion of War Aims’,

Security Studies, Vol. 6, No. 4 (1997), pp. 1–49, at pp. 4–5, 11; John Mearsheimer,Tragedy of Great Power Politics, pp. 2–3, 21. I expose this implicit conclusion or assump-tion of offensive realism in Shiping Tang, ‘Fear in International Politics’.

228 Tang Shiping


at Fudan U



ownloaded from


realism has been over-pressing their case against the possibility of coopera-

tion under anarchy.

Defensive Realists: Wrong and Indefensible Defence

Because proponents of ODT—many of whom are defensive realists—have

so far failed to grasp the valid level (and less often, utility) of differentiation

and the real challenge that the offensive realist attack against differentiation

poses, they have waged a wrong and indefensible defence against it.

Lynn-Jones, for instance, ignores the all-too-evident political motivation

behind the offensive realist attack on differentiation. Although repeatedly

pointing out that even opponents of differentiation (mostly offensive real-

ists) consistently differentiate military capabilities, postures and strategies,72

he is still not prepared to stand up for differentiation of military posture,

being more interested in defending the validity of ODB.73

Goldfischer recognizes that during the Cold War years many Cold War

hawks or offensive realists denied the possibility of differentiating military

postures or weapons, either implicitly or explicitly for reasons of arguing

against arms control.74 Believing it would be dangerous for the United

States to pursue any kind of cooperation—of which arms control is one

form—with the Soviet Union,75 they maintained that the logic of arms

control—which critically depends on the differentiation of weapons—was

fundamentally flawed and that any attempt at arms control would hence

fail. Goldfischer, however, chooses the indefensible defence of weapons

differentiation.

Despite their mistaken convictions as regards the possibility of cooper-

ation under anarchy, offensive realists have been more correct on this point.

Goldfischer, heavily influenced as he is by the intellectual legacy of arms

control and eager for a technological solution to the problem of war and

pace, could not be more wrong.

First, arms control, especially qualitative arms control,76 depends on the

distinguishability of weapons. It is otherwise difficult to determine what

types of weapons should be limited, scrapped, or banned.

72 Sean Lynn-Jones, ‘Offense-Defense Theory and Its Critics’, pp. 672–676, footnotes 24, 27,35, 41, and 44.

73 Lynn-Jones might also have a selfish interest in doing so, as ignoring politics allows him tostick to his thesis that war is the continuation of technology, not politics. On this point,offensive realists know better. See the section on the ODB for details.

74 David Goldfischer, The Best Defense, pp. 16, 20.75 See, for example, Colin Gray, Weapons Don’t Make War, p. 64; Samuel Huntington, ‘U.S.

Defense Strategy’, p. 42; and Albert Wohlstetter, ‘The Political and Military Aims ofOffense and Defense Innovation’, in Fred S. Hoffman, Albert Wohlstetter, and DavidS. Yost, eds., Swords and Shields (Lexington: D. C. Heath, 1987), pp. 5–6.

76 I thank Evan Montgomery for reminding me about the distinction between qualitative andquantitative arms control. Of course, even quantitative arms control may involve somekind of classification of weapons.



at Fudan U



ownloaded from


Second, defensive realists have misconstrued the relationship between

arms control and cooperation-building by inverting it. Most advocates of

arms control believe that arms control causes cooperative relationship, thus

essentially calling for a technological fix to adversarial relationships in inter-

national politics.77 Arms control, however, is more likely to be the outcome

than the cause of a cooperative relationship. In a confrontational relation-

ship between two states an arms race is probable, but in a cooperative

relationship arms reduction or control is far more likely, and each of the

two is less concerned about the other’s arms build-up.78 Moreover, arms

control is not the best start to a cooperative relationship between two states,

because making such demands is more likely to impede than promote trust

and cooperation.79

By defending the differentiation of weapons, proponents of ODT have

waged a wrong and indefensible defence against the offensive realist

attack on differentiation. The correct and defensible defence is that as

long as military postures can be differentiated, signalling benign intentions

through military postures within an acceptable level of risk will be pos-

sible.80 This is where the real utility of differentiating military postures lies.

The Objective ODB

The ODB can be unpacked into two versions: objective and subjective, each

of which has quite different implications: The objective balance influences the

outcome of battle and war; the subjective balance influences the cause and the

timing of war and the pattern of alliance pattern before the war breaks out.81

In other words, the objective balance influences the outcomes from the

interaction of state’ behaviour; whereas the subjective balance influences

state behaviour itself.

So far, both proponents and opponents of ODT have either failed to make

an explicit differentiation of these two versions of ODB,82 or more

77 Robert Jervis, ‘Cooperation under the Security Dilemma’, pp. 199–201; Charles Glaser,‘Political Consequences of Military Strategy’, Charles Glaser, ‘Realists as Optimists’,pp. 64–7; and Stephen Van Evera Causes of War, pp. 190–192.

78 Colin Gray makes a similar point, but over-states it. He argues, ‘you only get arms controlwhen you don’t actually need it’. See Colin Gray, Houses of Cards: Why Arms ControlMust Fail (Ithaca, NY: Cornell University Press 1992). In my A Theory of SecurityStrategy, Chapter 5, I argue in detail why some forms of arms control can be achievedwhen needed (i.e. it can facilitate cooperation).

79 Shiping Tang, A Theory of Security Strategy, Chapter 5.80 Charles Glaser, ‘Realists as Optimists’, pp. 68–70; Robert Jervis ‘Cooperation under the

Security Dilemma’, pp. 199–206; Andrew Kydd, ‘Sheep in Sheep’s Clothing’, pp. 143–7;and Evan Montgomery, ‘Breaking out of the Security Dilemma’, p. 163.

81 Note, however, that the ex ante alliance pattern also influences states’ strategies, and hencemay in turn influence states’ decisions on offence and defence. See Scott Sagan, ‘1914Revisited’; Richard Betts ‘Must War Find a Way?’ pp. 194–5.

82 See, for example, Karen Adams, ‘Attack and Conquer?’; Bernard Finel, ‘Taking Offense’;Stace Goddard, ‘Taking Offense’; and Stephen Van Evera, ‘Correspondence: Taking

230 Tang Shiping


at Fudan U



ownloaded from


commonly, differentiated them but failed explicitly to acknowledge their

different implications.83 Within the existing literature, only Biddle, Lieber

and Tang have clearly grasped the different implications of objective and

subjective ODB,84 but their elaborations remain incomplete (see below).

Understanding of the balance and its implications is consequently confused.

This section examines objective ODB, which can be further unpacked into

two sub-versions. They are: technology-only and technology-plus.85

The Technology-only Version of the Objective Balance:Impossible

The technology-only version of the objective balance is a systemic or struc-

tural variable on which to develop a structural theory on the balance and

outcome of war.86 The problem is that it is impossible to operate.

Two pre-conditions are necessary to make a valid examination of the

impact a particular weapon or military technology has on the objective

balance. The first is that both belligerents in a war must possess either the

weapon or the technology.87 The second is that states possessing the same

technology must, or must be assumed to deploy it optimally.88 Essentially,

when measuring the technology-only version of the objective balance, the

Offense at Offense-defense Theory’. Karen Adams expands the ODB into offence-defence-deterrence balance, making the balance even more complex. Because the generallogic here also applies to her formulation, I do not discuss offence-defence-deterrencebalance separately.

83 See, for example, Richard Betts, ‘Must War Find a Way?’, p. 180, footnote 17; ThomasChristensen and Jack Snyder, ‘Chain Gangs and Passed Bucks’; James Davis, ‘TakingOffense’, pp. 180–1; James Fearon, ‘The Offense-Defense Balance and War since 1648’;Charles Glaser and Chaim Kaufmann, ‘What Is the Offense-Defense Balance?’; YoavGortzak et al, ‘Offense-Defense Theory’; Jack Levy, ‘The Offensive/Defensive Balanceof Military Technology’; Sean Lynn-Jones, ‘Offense-Defense Theory and Its Critics’,pp. 666, 670–1, 679–82, footnote 15; Robert Jervis ‘Cooperation under the SecurityDilemma’, pp. 190–4; and Stephen Van Evera, ‘Correspondence: Taking Offense atOffense-defense Theory’. Within the existing literature, only Biddle, Lieber and Tanghave clearly grasped the different implications of the two versions of the balance.

84 Stephen Biddle ‘Rebuilding the Foundation of Offense-Defense Theory’, pp. 744–5; KierLieber, ‘Grasping the Technological Peace’, pp. 73, 102–3; Kier Lieber, War and theEngineers, pp. 3–4, 24; and Shiping Tang, ‘A Systemic Theory of the SecurityEnvironment’, pp. 7–8, 28–9.

85 Charles Glaser and Chaim Kaufmann label these two versions as ‘narrow’ and ‘broad’,whereas Lieber labels them as ‘core’ and ‘broad’. See Charles Glaser and ChaimKaufmann, ‘What Is the Offense-Defense Balance?’; Kier Lieber, ‘Grasping theTechnological Peace’; Kier Lieber, War and the Engineers.

86 Kier Lieber, ‘Grasping the Technological Peace’, pp. 76–7; and Sean Lynn-Jones,‘Offense-Defense Theory and Its Critics’, p. 668.

87 This point, which should be obvious, is often missing from the existing discussion, or onlyimplicitly spelled out. See, for example, James Fearon, ‘The Offense-Defense Balance andWar since 1648’, p. 6; and Kier Lieber, ‘Grasping the Technological Peace’, p. 75.

88 James Fearon, ‘The Offense-Defense Balance and War since 1648’, pp. 7–8; StaceGoddard, ‘Taking Offense’, pp. 192–4; and Kier Lieber, ‘Grasping the TechnologicalPeace’, p. 75. Indeed, without recognizing how demanding this optimality assumptionis, Glaser and Kaufmann, both of whom are proponents of ODB, actually employ theoptimality assumption to measure the objective balance, even though they favour a broad



at Fudan U



ownloaded from


human factor in the two sides to a (potential) conflict must be assumed absent,

and technologies or weapons to operate independently of human decisions.89 It

is otherwise impossible to state with confidence that a particular technology

confers a distinct advantage on either defence or offence. Once we acknow-

ledge that the technology-only version of the objective balance works only

under these two stringent conditions, it becomes evident that it is impossible

to operate.

To begin with, the technology-only version of the objective balance can be

defined as a relative advantage that technology or several technologies

confer on offence or defence, and nothing more. Any definition that goes

beyond this is not a valid definition of the technology-only version of the

objective balance.

Both Lynn-Jones and Lieber favour the technology-only objective bal-

ance, but neither makes an accurate definition of it.90 It is, according to

Lynn-Jones, ‘the amount of resources that a state must invest in offence to

offset an adversary’s investment in defence’,91 which implies the need for

state decisions or input from human factors, and does not, therefore, assume

optimal deployment of technology. Lieber defines the technology-only bal-

ance as ‘some measures of the relative easy of attacking and taking territory

versus defending territory’,—a definition that implies the need for a state to

make the strategic decision of attacking to occupy territory which, again,

does not rule out the human factor.92

Second, the first condition makes it apparent that some of the supposed

impact of technology upon the technology-only version of the objective

ODB is not really the technology’s impact on the ODB per se, but rather

the impact of possessing a lead in technological innovation upon the out-

come of war (at different levels). Many of the cases that ODT proponents

employ to advance the thesis that conquest is easy when offence is dominant

simply reflect the impact upon the outcome of war (at different levels) of

possessing a lead in technological innovation, rather than of effecting a

change in the technology-only version of the objective ODB per se.93

(i.e. technology-plus) approach. See Charles Glaser and Chaim Kaufmann, ‘What Is theOffense-Defense Balance?’ p. 201.

89 Taylor Fravel raises the possibility that the human factor might have been assumed absentfor the sake of theory-building by proponents of ODT. This defence is similar to JackSnyder’s overall defence of ODT, and thus ultimately untenable (see below).

90 Karen Adams also seeks to measure a technology-only version of the objective balance. Idiscuss her work in detail below.

91 Sean Lynn-Jones, ‘Offense-Defense Theory and Its Critics’, p. 665.92 Kier Lieber, ‘Grasping the Technological Peace’, p. 74. As Finel points out, taking terri-

tory is not and should not always be the goal of offences if the goal is to win a war.Bernard Finel, ‘Taking Offense’, pp. 184–5. Lieber later changes his definition to ‘therelative ease of attack and defense given prevailing conditions’. See Kier Lieber, Warand the Engineers, p. 27. As this is even broader than his original definition, it is againnot a technology-only definition.

93 Both Fearon and Lieber argue that when one state achieves a technological innovation, itrepresents a change in the balance of power between the state and its potential adversaries,

232 Tang Shiping


at Fudan U



ownloaded from


For instance, guns became suitable for offensive warfare after the break-

through under Charles VIII in 1494 that made guns standardized, mobile,

and easy to fire.94 But this does not mean that guns in themselves conferred

unique advantages on offence when taking into account what would have

happened if Charles VIII’s enemies had possessed the same type of gun.

Similarly, although the blitzkrieg would have been difficult without

tanks,95 this does not mean that tanks confer a distinctive advantage on

offence. Speed, by allowing defence rapidly to respond to a breaching of

frontal defences and to close the gap, also confers a distinctive advantage

upon defence. As Bernard Brodie says, ‘. . .if the French had disposed of a

properly concentrated armoured reserve, it would have provided the best

means for their cutting off the penetration and turning into a disaster for the

Germans what became instead an overwhelming victory.’96 Hitler’s panzer

armies initially scored successes until their enemies, notably the Soviet

Union, devised effective countermeasures to cut off panzer army advances

and prevented their deep penetration.97

Understood correctly therefore, many so-called periods of offence or de-

fence dominance were due to the simple fact that a military innovator—a

state or a general that developed a new technology or a new approach to

deploying existing technologies—enjoyed the typical first-mover’s advan-

tage, and had nothing to do with the technology-only version of the object-

ive balance.98 Because most, if not all, military innovations—whether in

technology, deployment of technology, tactics, organization, or doctrine—

are not available to every state (otherwise, they would not be called ‘innov-

ations’), determining whether an innovation confers inherent advantage on

offence or defence is impossible. Without even acknowledging that the ob-

jective balance requires both sides of a conflict to possess the same technol-

ogy, proponents of ODB critically weaken their case by repeatedly talking

about technological ‘innovations’, ‘advances’, and ‘improvements’ rather

than technology per se.99

not a change in ODB per se. While the first half of their assertion may not be true, thesecond half is. James Fearon, ‘The Offense-Defense Balance and War since 1648’, p. 7; andKier Lieber, ‘Grasping the Technological Peace’, p. 76.

94 John Keegan, A History of Warfare (London: Verson, 1993), pp. 320–2.95 Sean Lynn-Jones, ‘Offense-defense Theory and its Critics’, p. 676.96 Bernard Brodie, War and Politics (London: Longman, 1974), p. 325.97 Kier Lieber, ‘Grasping the Technological Peace’, pp. 90–4; and Jack Snyder, ‘Perceptions

of the Security Dilemma in 1914’, in Robert Jervis, Richard Ned Lebow and Janice Stein,eds., Psychology and Deterrence (Baltimore: John Hopkins University Press, 1985),pp. 159–60.

98 Jack Levy, ‘The Offensive/Defensive Balance of Military Technology’, pp. 230–4. A newand more effective countermeasure against an existing technology is another form of in-novation, and should hence again confer first-mover’s advantage on the innovator.

99 See, for example, Charles Glaser and Chaim Kaufman, ‘What is Offense-defenseBalance?’, pp. 62–6; and Stephen Van Evera, ‘Correspondence: Taking Offense atOffense-defense Theory’, pp. 195–6.



at Fudan U



ownloaded from


Third, the second condition means that even if the same technology were

available to more than one actor, different actors might deploy it in different

ways, usually for reasons other than technology. This makes assessing the

impact of technology upon offence or defence even more difficult.

For instance, before WWII, tanks were available to both Germany on the

one side and France and Britain on the other. But because Hitler wanted

expansion and conquest, Germany deployed tanks for offensive purposes,

whereas France and Britain, because their purpose was not expansion and

conquest, at least not on the European Continent, deployed tanks primarily

for defensive purposes. So, does the fact that Hitler’s panzer armies initially

scored spectacular successes mean that tanks are best deployed as offensive

weapons? Hardly so; when in retreat and on the defensive, Germany also

deployed tanks as effective weapons for slowing down the allied advance

into Germany’s heartland.100

Fundamentally, the optimality assumption means that the technology-

only version of the objective ODB can only be ex post and ad hoc justifica-

tions of what happened in history. One form of deployment of a technology

produced a good outcome, and that particular form of technology deploy-

ment is hence considered as ‘optimal’.101 But any outcome of wars and

battles is surely the result of interaction of states’ strategic, tactical, and

operational decisions, with technology playing only a small role in shaping

the outcome.102 More often than not, one side’s good outcome in wars and

battle is due to the fact that the other side made mistakes or simply did not

have the power necessary to win.

Indeed, without actually grasping the devastating effect on their optimal-

ity assumption of their qualification for measuring objective (technology-

plus) ODB, Glaser and Kaufman say that ‘the effects of innovation in

protection, logistics, communication, and detection are more varied,

depending on specific interaction with force behaviour’.103 This statement

is an admission that the optimality assumption usually cannot be operated,

and hence nullifies much of their logic for the (technology-plus) objective

ODB.

The human factor in war cannot be totally eliminated. As a result, it is

either difficult, if not impossible, to assess the impact of a particular tech-

nology on either defence or offence, unless dealing with situations in which

the human factor is truly marginal (e.g. distance between states).104

100 Kier Lieber, ‘Grasping the Technological Peace’, pp. 92–3.101 Bernard Finel, ‘Taking Offense’, pp. 185–7.102 Ibid, pp. 183–4; and Kier Lieber, ‘Grasping the Technological Peace’.103 Charles Glaser and Chaim Kaufmann, ‘What Is the Offense-Defense Balance’, p. 64.

Measuring the technology-plus version of ODB will be more difficult than measuringthe technology-only version of ODB. See below.

104 Nationalism may be another factor that confers distinctive advantage on defense, butnationalism is not technology.

234 Tang Shiping


at Fudan U



ownloaded from


The technology-only version of the objective balance, therefore, despite its

aura of being both structural and parsimonious, is impossible to measure

and has little value for understanding the outcome of war or battle.105

This is inevitable, because in human society in which human ingenuity

rules, a purely technological explanation for the outcome of war cannot

be correct.

The Technology-plus Version of the Objective Balance:Intractable

Because the technology-only version of the objective balance can be of little

value for understanding the outcome of war, bringing in the human factor—

or taking a technology-plus approach to the objective balance—becomes

inevitable if the balance is to remain relevant.

But once we adopt a technology-plus approach, the objective balance

becomes profoundly complex, if not totally intractable. Proponents of the

objective balance are open to the possibility that different factors, including

weapons and technologies, might bring different benefits to offence or de-

fence—however contentious the argument itself may be—as signifiers that

the balance can be measured.106 Such a stand is simply wrong.

For instance, Glaser and Kaufmann raise the five major factors they be-

lieve influence the balance and which should hence be incorporated into the

technology-plus version of the objective balance. They are: technology,

geography, force dimension, nationalism, and cumulative resources.107

Each of these factors, they argue, can be broken down into several

sub-dimensions. For example, geography encompasses cover, movement

inhibition and distance. Technology includes mobility, fire-power, protec-

tion, logistics, communication, and detection.

The problem is that even if Glaser and Kaufmann were to be correct, and

these factors do have different influence on offence and defence, this does

not mean that pooling their impact on the objective balance enables the

objective balance to be measured. Intuitively, the exact opposite must be

true; it is precisely because so many factors influence offence or defence that

105 Lynn-Jones, Lieber, and Tang all favour the technology-only approach almost exclusivelyon the grounds of retaining its structural and parsimonious nature. Sean Lynn-Jones,‘Offense-defense Theory’, p. 666, footnote 15; Kier Lieber, ‘Grasping TechnologicalPeace’, pp. 75–7; Shiping Tang, ‘A Systemic Theory of the Security Environment’,pp. 7–8. Earlier, Fearon and Levy also question the feasibility of a structural (i.e.system-wide) ODB. See James Fearon, ‘The Offense-Defense Balance and War since1648’, pp. 12–13; Jack Levy, ‘The Offensive/Defensive Balance of Military Technology’,p. 227.

106 See, for example, Charles Glaser and Chaim Kaufmann, ‘What Is the Offense-DefenseBalance?’; Stephen Van Evera, ‘Correspondence: Taking Offense at Offense-defenseTheory’, pp. 195–6; Karen Adams, ‘Attack and Conquer?’

107 Charles Glaser and Chaim Kaufmann, ‘What Is the Offense-Defense Balance’, pp. 60–8.Glaser and Kaufmann rightly exclude alliance behaviour and first-move advantages fromtheir definition of the broader version of objective ODB.



at Fudan U



ownloaded from


it will be extremely difficult to measure the balance, even through the ‘net

assessment’ approach that these two authors propose.

Take mobility (under technology) for example. Glaser and Kaufmann

break mobility down into operational, tactical, and strategic mobility.108

But even if proponents of the technology-plus version of the objective bal-

ance were to agree that operational mobility, tactical mobility, and strategic

mobility all favour either offence or defence—a matter upon which propon-

ents of objective ODB have yet to agree, judging from the Glaser and

Kaufmann discussion—wouldn’t piling up different types of mobility

make measuring the impact of mobility upon offence and defence an ex-

tremely difficult task?109 And when the objective balance includes factors

other than just mobility, ‘a reader would need a cue card just to be able to

think about how to apply it [i.e. Glaser and Kaufmann’s measurement of the

balance]’.110

In light of the preceding discussion, it becomes evident that Van Evera’s

broader approach, which takes balancing behaviour, power, and diplomacy,

among other things, as part of the balance, is also not valid. Van Evera’s

definition of offence dominance as ‘when conquest is fairly easy’111 is flawed,

because it ‘conflates offence dominance with a host of other variables’, thus

making a grab bag of the (objective) balance.112 By any measure, Van Evera

has made his version of the ODB equivalent to the whole realist theory of

war, or at least made ODB practically indistinguishable from relative

power.113 Van Evera simply includes other realist factors or explanations

of war in the objective balance, re-packages them, and claims that the (ob-

jective? subjective?) balance is now the ‘master key’ to understanding war.114

Unfortunately, his whole theoretical enterprise is built on sand; his defin-

ition is flawed, his logic is tautological, and his testing is woefully inadequate

(see below).115

108 Ibid., pp. 62–3.109 For a summary of the differences among proponents of ODT regarding the impact of

mobility on offense or defence, see Kier Lieber, War and the Engineers, pp. 35–42.110 Richard Betts, ‘Must War Find a Way?’ p. 186.111 Stephen Van Evera, Causes of War, p. 118, footnote 2.112 Stace Goddard, ‘Taking Offense’, p. 190.

offence-defence theory: towards a definitive understanding€¦ · pp. 108–146; stephen van...

Documents